1. Field of the Invention
The present invention generally concerns CMOS power amplifiers, particularly as are operated class AB.
The present invention particularly concerns CMOS power amplifier circuit design, particularly as operated class AB and particularly as employ an NMOS input transistor, canceling at least one circuit nonlinearity.
2. Background of the Invention
2.1 General Background of the Invention
In a wireless communications system the linearity and efficiency of the system power amplifier(s) directly limits the performance of the entire system. The efficiency of the power amplifier is critical for extending the battery lifetime, while its linearity is necessary for the transmission of a quality digital signal.
The Class AB power amplifier is widely used in wireless communications systems because it is relatively easy to design and build, and has fairly good linearity and efficiency. Most recently, circa 2001, CMOS RF power amplifiers are receiving more and more attention because these amplifiers are more cost effective than are amplifiers constructed from GaAs and silicon bipolar transistors while being suitably integrated with other CMOS IC parts, primarily digital logic. Understanding the sources of any nonlinearity in NMOS Class AB amplifier is useful for improving its performance.
The linearity of CMOS Class AB power amplifier is mostly limited by two sources. The first source is the nonlinearity of the amplifier input capacitance, which can be modeled as (1) a gate-to-drain capacitance Cgd in electrical series with a CMOS transistor selectively conducting current Id, plus (2) a gate-to-source capacitance Cgs in electrical parallel with the same CMOS transistor. (To complete the model, a drain-to-source resistance Rds, and a drain-to-source capacitance Cds are also, further, in electrical parallel with the CMOS transistor.)
When the CMOS transistor toggles on and off during Class AB operation, its gate capacitance Cgs changes dramatically. This varying capacitance will cause a nonlinearity at the gate. In other words, the gate voltage Vgd will be a nonlinear version of the input signal because of this capacitance.
This equivalent circuit of a CMOS transistor at the signal input of a class AB CMOS power amplifier may be simulated and analyzed, with, by way of example, SPICE, for the actual parameters of various CMOS technologies, for example the Mosis HP CMOS 0.6 pm technology. For the technology Cgs roughly changes from 1 pF/mm to 0 pF/mm as the CMOS transistor is turned on and off.
A second source of nonlinearity is the nonlinear transconductance Gm of the CMOS transistor. When the CMOS transistor is switching on and off, its drain current is also switching on and off, thus making the transconductance Gm to be a nonlinear function of the transistor gate voltage Vgs. The present invention will show how to minimize this second nonlinearity.
These non-linearities are not only of considerable percentage difference, they are noticeable in the circuit response of a high-power class AB amplifier. As requirements for driving lower and lower (speaker impedance) have arisen (i.e., 32 ohms goes to 4 ohms), while power FETs grow larger and larger, the parasitic capacitance of these FETs Cgs continues to grow.
In view of the foregoing, a need thus exists for a FET, and particularly a CMOS FET power amplifier that manages these effects better while remaining capable and efficient of high drive capability (and, likely also, low quiescent current drain).
2.2 Specific Prior Art to the Invention
U.S. Pat. No. 4,315,223 to Haque for an OPERATIONAL AMPLIFIER WITH IMPROVED FREQUENCY COMPENSATION is of relevance to the present invention for showing that the effects of circuit capacitance may be compensated. The Haque patent concerns an operational amplifier circuit (i.e., not a class AB amplifier circuit) comprised of complementary MOS transistors with a bias section, a differential amplifier section, a level shift stage and an output stage. The circuit provides for frequency compensation using two capacitors. One capacitor, connected between the differential amplifier section and the output stage through a CMOS transmission gate that functions as a resistor, acts as the dominant pole of the transfer function. A second capacitor between the amplifier section output node and a level shift transistor, functions to remove the secondary poles in the transfer function and cause the dominant pole to occur at a higher frequency.
U.S. Pat. No. 4,988,902 to Dingwall for a SEMICONDUCTOR TRANSMISSION GATE WITH CAPACITANCE COMPENSATION is if relevance to the present invention for showing that a semiconductor elementxe2x80x94herein a transfer gatexe2x80x94may be used to supply a capacitance that is used in a circuit (not an amplifier circuit) for purposes of capacitance compensation. The Dingwall patent shows a transmission gate employing a pair of capacitors ahead of and a pair of capacitors behind a transistor. One capacitor of each pair is supplied with a control voltage pulse that leads and the other with a control voltage pulse that lags the complement of a control voltage pulse supplied to the gate of the transistor. The capacitors are typically each a MOS transistor with the gate serving as one terminal and the drain and source shorted together and serving as the other terminal. Moreover, each of the capacitors has a capacitance equal to one half the capacitance of the gate to source and gate to drain capacitance of the transistor. This circuitry makes possible charge compensation to avoid the build up of trapped charge in the transistor. The capacitance of the pair of capacitors ahead of the transistor is approximately equal to the gate-to-drain parasitic of the transistor and the capacitance of the pair of capacitors behind the transistor is equal to the parasitic capacitance of the gate-to-source of the transistor.
U.S. Pat. No. 5,148,065 to Krenik for CURRENT STEERING CIRCUITS AND METHODS WITH REDUCED CAPACITIVE EFFECTS concerns capacitance compensation techniques used to reduce capacitive effects that impact on the performance of certain current steering circuits. The circuits dealt with by Krenikxe2x80x94switching circuitsxe2x80x94are not equivalent to the class AB amplifier circuit of the present invention. The techniques of Krenik in modifying the physical layout and structure of components, mainly transistors, used to implement a current steering circuit, are different than the compensation/cancellation technique of the present invention will be seen to be. Nonetheless, Krenik shows how capacitive effects arise in semiconductor, including CMOS, circuitry, and the limitations of semiconductor technology in compensation for these capacitive effects.
In one isolation technique of Krenik a resistor (R) or a diode (D) is coupled to a data-switched transistor to dampen voltage perturbations associated with the gate-to-source capacitance. In another technique a transistor (PDV) is included in either the output or ground legs of the current steering circuit to provide a design variable to counteract the capacitive effects of the associated data-switched (PDX/NDX) or voltage-controlled (PREF) transistor. In a bipolar substitution technique a data-switched bipolar transistor (QDX) is substituted for the data-switched MOS transistor, and made sufficiently small to significantly reduce junction capacitance. In addition, capacitive effects can be reduced by introducing fabrication alterations such as fabrication layouts in which the transistor""s source contact is made within a U-shaped gate, and in which the transistor""s moat perimeter is contoured for minimal gate area consistent with standard gate/contact spacing requirements.
U.S. Pat. No. 5,635,873 to Thrower, et al., for an OPERATIONAL AMPLIFIER HAVING DUMMY ISOLATION STAGE is of remote relevance to the present invention for showing that, rather than dealing with parasitic capacitance of a circuit, an attempt can be madexe2x80x94at least in a bipolar transistor circuitxe2x80x94to deal with circuit mismatchxe2x80x94presuming as may arise from parasitic capacitance, although any particular association of the parasitic capacitance with the gate capacitance of a CMOS transistor is likely a somewhat remote, and strained, associationxe2x80x94to deal with the currents produced by the circuits.
The Thrower, et al., patent concerns an integrated circuit terminator for a SCSI bus with resistors made of laser-blowable fuses in an array and a reference voltage source made with a bandgap generator and a two stage amplifier including a dummy isolation stage for providing symmetrical mismatch currents. In an amplifier having (i) a differential input stage having a number of transistors arranged in a cascode configuration; (ii) an isolation stage coupled to this differential input stage, this isolation stage with first and second current sources in series and including at least one bipolar junction transistor; and (iii) a dummy stage coupled to said differential input stage symmetrically with respect to said isolation stage, the dummy stage with third and fourth current sources in series, the third current source matching said first current source and said fourth current source matching said second current source, the dummy stage is distinguished by further including at least one bipolar junction transistor. By this configuration current source mismatches within said isolation stage between the first current source and the third current source are replicated in the dummy stage and symmetrically applied to the differential input stage.
U.S. Pat. No. 6,255,909 to Muza for an ULTRA LOW VOLTAGE CMOS CLASS AB POWER AMPLIFIER WITH PARASITIC CAPACITANCE INTERNAL COMPENSATION is perhaps the patent of closest relevance to the present invention although, in dealing with the output transistors of a class AB amplifier, it presents a technique (i) non-analogous to the technique of the present invention (ii) operative at the other side of an AB amplifier from the input stage where the present invention will be seen to be operative. The Muza patent is relevant because it shows that parasitic capacitance is of relevance in the operation of both CMOS circuitry and, in particular, a class AB amplifier.
In the Muza patent an ultra low voltage CMOS, class AB power amplifier has internal compensation using only parasitic gate capacitance. The class AB power amplifier has (i) an input gain stage, (ii) a PMOS output transistor having its source connected to a first supply voltage Vdd and having its gate driven by a PMOS driver transistor in response to a positive input signal generated via the input gain stage, and, further (iii) a NMOS output transistor having its source connected to a second supply voltage Vss and having its gate driven by a NMOS driver transistor in response to a negative input signal generated via the input gain stage. The output transistors and the driver transistors are configured as a class AB output stage. Moreover, the PMOS output transistor and the NMOS output transistor are selected to have predetermined gate-source parasitic capacitance values. This selection is such that the gate-source parasitic capacitance of the PMOS output transistor and the gate-source parasitic capacitance of the NMOS output transistor combine to achieve a class AB power amplifier having a single dominant pole. This techniques is usable where the input gain stage and the output stage are devoid of Miller compensation. The class AB power amplifier so implemented substantially eliminates unwanted effects of parasitic capacitance associated with the output transistors.
The present invention contemplates a new and nonlinear cancellation technique developed specifically for NMOS power amplifiers operated in class AB. This technique applies a PMOS transistor to compensate for changes of the gate capacitance of an NMOS input transistor during the switching on and off of this transistor during Class AB operation of the circuit of which the transistor is a part, thus improving the linearity of the whole circuit.
1. An Improved Class AB CMOS Transistor Amplifier
In its principle aspect, the present invention is embodied in an improved class AB CMOS transistor amplifier.
The improved amplifier is based on a conventional, preexisting, class AB amplifier receiving an input signal at a gate of one of a NMOS or a PMOS type CMOS transistor, with this gate of the CMOS transistor having a nonlinear gate capacitance.
To this amplifier is added another, opposite-type, one of a NMOS or a PMOS type CMOS transistor. This added transistor also receives at its gate the same input signal, and has a nonlinear gate capacitance that is essentially opposite to the nonlinear gate capacitance of the existing CMOS input transistor of the class AB amplifier (the only transistor actually to amplify the input signal, and to send it on).
The nonlinear gate capacitance of this added, opposite type, CMOS transistor serves to compensate for the nonlinear gate capacitance of the normal and pre-existing CMOS transistor of the class AB amplifier, thus reducing any nonlinearity of the class AB amplifier which is due to the nonlinear gate capacitance of its CMOS transistor receiving the input signal. This added, opposite-type, CMOS transistor thus serves as nonlinear-gate-capacitance compensation transistor used with the class AB amplifier.
In one variant embodiment a class AB amplifier receives the input signal at the gate of a NMOS type CMOS transistor having an nonlinear gate-source capacitance Cgs1. To this circuit structure is added a PMOS type CMOS transistor that also receives at its gate the same input signal, thereby with its nonlinear gate source capacitance Cgs2 essentially cancel the nonlinear gate capacitance Cgs1 of the NMOS type CMOS transistor of the class AB amplifier.
It is clearly possible to have an alternative variant embodiment of the opposite sense, where a NMOS type transistor is added to compensate for variations in the gate capacitance of an input PMOS transistor to a class AB amplifier.
In the preferred embodiment the first-type CMOS transistor of the input amplifier is biased to turn on substantially exactly when the opposite second-type CMOS compensation transistor turns off, and vice versa.
2 A Method of Compensating for a Non-Linear Gate Capacitance at the Gate of NMOS- or a PMOS-Type CMOS Transistor of a Class AB Amplifier
In another of its aspects, the present invention is embodied in a method of compensating a non-linear gate capacitance at the gate of NMOS- or a PMOS-type CMOS transistor of a class AB amplifier.
In the method an input signal is first-received at a gate of one of a NMOS- or a PMOS-type CMOS first transistor of a class AB amplifier. This gate has a nonlinear gate capacitance.
Meanwhile, the same input signal is second received at another NMOS- or PMOS-type CMOS second transistor of opposite type to the first transistor. This second transistor itself has a nonlinear gate capacitance that is essentially opposite to the nonlinear gate capacitance of the first transistor.
The nonlinear gate capacitance of the other, opposite type, CMOS second transistor serves to compensate for the nonlinear gate capacitance of the CMOS first transistor of the class AB amplifier. Any nonlinearity of the class AB amplifier as is due to the nonlinear gate capacitance of its CMOS first transistor that receives the input signal is thus reduced. The opposite type, CMOS second transistor thus serves as a nonlinear-gate-capacitance compensation transistor for use with a class AB amplifier.
In the method the first-receiving of the input signal may for example be at the gate of a NMOS-type CMOS input transistor of the class AB amplifier, this NMOS-type CMOS transistor having an nonlinear gate-source capacitance Cgs1, while the second-receiving of the input signal is at the gate of PMOS-type CMOS transistor having a nonlinear gate source capacitance Cgs2 that essentially cancels the nonlinear gate capacitance Cgs1.
In the method (i) the first-type CMOS first transistor of the input amplifier is preferably biased to turn on consequent to this first transistor""s first-receiving of the input signal substantially exactly when (ii) the opposite second-type CMOS compensation second transistor is biased to turn off consequent to this second transistor""s second-receiving of the input signal, and vice versa.
These and other aspects and attributes of the present invention will become increasingly clear upon reference to the following drawings and accompanying specification.